Distributed fault detection based on HMM for Wireless Sensor Networks

In their practical applications, Wireless Sensor Networks (WSNs) can experience faults during running. Furthermore, faulty data collected in the base station might lead to a wrong judgement. Since the whole system´s performance require the accuracy of data, node fault detection is necessary. Motivated by the requirement of fault detection performance, we propose a distributed fault detection algorithm for wireless senor networks based on Hidden Markov Model (HMM). Firstly, a density of probability method is presented to obtain the possibility of translation of state for each node in the network. Then, a HMM is introduced to characterize the tendency of each node to be faulty. Finally, a distributed fault detection algorithm is proposed based on the HMM. Simulation results show that the proposed HMM-based fault detection achieves high detection accuracy.